Risk Factors and Outcomes for Multidrug-resistant Gram-negative Bacterial Infection in Adult Patients with Abdominal Surgery Requiring Intensive Care

Background: Multidrug-resistant (MDR) gram-negative bacterial (GNB) infections remain a signicant cause of morbidity and mortality among surgical patients. The objective of our study was to recognize the risk factors for MDR GNB infection in surgical intensive care unit (SICU) patients with abdominal surgery and determine the predictors independently associated with death. Methods: From 2010 to 2017, a retrospective cohort study was conducted among patients with abdominal surgery admitted in SICU. Patients with GNB (MDR and non-MDR) infections were included for analyses. Results: A total of 364 patients with abdominal surgery experienced GNB infections, among them, 117 (32.1%) were MDR GNB infections. Of 133 MDR GNB isolates, the most frequent isolate was Escherichia coli (45.1%). Patients with MDR GNB infection had signicantly longer ventilator days and hospital stay, as well as higher 30-day and in-hospital mortality compared to non-MDR GNB patients. Multivariable analysis showed longer length of pre-ICU stay, surgical re-exploration, received anti-pseudo carbapenems and uoroquinolones, and higher total bilirubin were independent risk factors for the acquisition of MDR GNB infection. Predictors for 30-day mortality among patients with MDR GNB infection were chronic kidney disease, received anti-pseudo carbapenems and inappropriate empirical antimicrobial therapy. Conclusions: This study provides important information about the risk factors for subsequent MDR GNB infection and 30-day mortality among the patients with abdominal surgery. Given the increasingly MDR GNB infection and mortality, implementation of antimicrobial stewardship programs in critical care unit is crucial to reduce MDR pathogens and optimize antimicrobial therapies.


Introduction
Despite the advances in infection control and postoperative care, bacterial infections remain a major problem after surgery and contribute signi cantly to the rate of morbidity and mortality [1][2][3]. Besides, postsurgical infection leads to increased length of hospital stay, higher rates of hospital readmission and higher healthcare costs [4][5][6]. While gram-negative bacteria (GNB) are becoming more common pathogens of postoperative infections, multidrug-resistance (MDR) strains among GNB were more prevalent substantially [7,8] with the incidence up to 48-65% in critically ill patients [9,10]. Previous studies have identi ed risk factors for acquiring MDR GNB infection [11,12]. Identifying a subgroup of critically ill patients who have a high risk of harboring MDR GNB infection after abdominal surgery would have important implications for patient care and outcomes. Meanwhile, antimicrobial stewardship is particularly important for intensive care unit (ICU) settings due to increasing trends in antimicrobial resistance of bacterial pathogens. The online antimicrobial-stewardship program has been implemented in the ICUs of the Kaohsiung Chang Gung Memorial Hospital, as previously reported [13,14]. Brie y, all antimicrobial agents prescribed to ICU patients required approval from infectious diseases physicians. If a prescription was disapproved, the antimicrobial would be discontinued within 48 hours, and the prescriber would be noti ed to modify the regimen [13,14]. In the present study, we sought to recognize the risk factors for the development of infection caused by MDR GNB in surgical ICU (SICU) patients who underwent abdominal surgery and determine the predictors independently associated with death.

Study design, setting and participants
From January 2010 to December 2017, all consecutive adult patients (≥ 18 years old) who underwent abdominal surgery and admission to SICU, at the Kaohsiung Chang Gung Memorial Hospital, a 2,600-bed primary care and tertiary referral medical center in Taiwan, were retrospectively included. The SICU is a 23-bed multispecialty intensive care center for critically ill patients who receive surgery including severe trauma and acute abdomen surgery (stomach, intestine/colon, appendix, rectum, liver, biliary system, pancreas, spleen and kidney), or those experiencing shock, cardiac arrest or sepsis after major surgery.
Patients were categorized as those either with MDR or non-MDR GNB infections. A comparison was made between MDR and non-MDR GNB patients to determine the independent risk factors for the acquisition of MDR GNB. To analyze the predictors of 30-day mortality among MDR GNB patients, we further compared clinical characteristics and laboratory ndings as well as complications of survivors and non-survivors.

De nitions
The surgical site was divided into (1) digestive system (includes esophagus, stomach, duodenal, small intestinal, colon, appendix and rectum) and (2) hepatobiliary/pancreas/spleen and kidney. If a patient had multiple episodes of GNB (MDR and non-MDR) infection during the study period, only the rst episode was included for analyses. In cases of polymicrobial infections, the episode was de ned as an MDR GNB case if one of the isolates was an MDR GNB strain. Chronic kidney disease is de ned as kidney damage or glomerular ltration rate < 60 mL/min/1.73 m2 for 3 months or more, irrespective of cause [15]. MDR is de ned as non-susceptibility to at least one agent in three or more antimicrobial categories according to the consensus de nition proposed by the Centers for Disease Control and

Data collection
The data were mainly retrieved from the Chang Gung Research Database, the largest multi-institutional electronic medical records collection in Taiwan [17], and were supplemented by a secondary manual search. The Chang Gung Research Database is a de-identi ed database derived from original medical records of Kaohsiung Chang Gung Memorial Hospital, with large volumes of database for research studies and analysis. We collected patient demographics and characteristics, Charlson Comorbidity Index, American Society of Anesthesiologists Classi cation, surgical wound classi cation [18], operative time, surgical sites, and number of re-explorations. We also retrieved the information regarding of invasive procedures and management including blood transfusion, surgical drainage (including pigtail, chest tube, Jackson-Pratt drains, abdominal open drains, and percutaneous transhepatic cholangiography and drainage), enteral tube (including nasogastric tube, nasoduodenal tube, jejunostomy tube, and gastrostomy tube), mechanical ventilation, invasive vascular access (including arterial lines, central venous catheters, Swan-Ganz catheters, large-bore catheters, and Hickman lines), foley catheter and hemodialysis. A medication information prescribed within 30 days before the onset of GNB infection was extracted including antibiotics, steroid, inotropic agents, chemotherapy and immunosuppressive agents.
Laboratory parameters included white cell blood platelet count hemoglobin creatinine C-reactive protein albumin alanine aminotransferase and total bilirubin levels were obtained. Microbiology specimens (including blood, ascites, bile, abscess, excision tissue, surgical wound and body uid) collection from different anatomical sites were obtained. The culture results of urine specimen and central venous catheter tip were excluded from the study. The length of stay prior to ICU admission, duration of mechanical ventilation, 30-day fatality after the onset of GNB infection and in-hospital mortality were record for analyses.

Antimicrobial susceptibility testing
Microbiology laboratories performed antimicrobial susceptibility testing of isolates using disk diffusion or automated testing methods according to guidelines and breakpoints established by the Clinical Laboratory and Standards Institute [19].

Statistical analysis
Values for continuous and categorical variables are expressed as means ± standard deviations (SD) and the number and percentage of the group from which they are derived. The ANOVA Wilcoxon test was used for continuous variables, and the chi-square test or Fisher's exact test was used for categorical variables, as appropriate. The logistic regression model with a stepwise procedure was used for multivariate analysis. Kaplan-Meier was used for 30-day survival analysis. All tests of signi cance used a 2-sided P < .05. Statistical analyses were conducted using SAS EG version 5.1.

Patients characteristics
A study ow-chart is shown in Fig. 1 MDR GNB) patients, 106 (29%) admitted directly from the emergency department to the SICU, and 258 (71%) patients were scheduled admissions subsequently required ICU care after surgery. Totally, 228 (62.6%) patients managed with emergency surgical intervention. Surgical re-exploration was found in 124 (34%) patients. The characteristic of the 364 patients with GNB (MDR and non-MDR) infection is summarized in Table 1. .002 Data were no (%) unless otherwise indicated. ALT alanine aminotransferase; ASA American Society of Anesthesiologists; CI con dence interval; CT chemotherapy; CRP C-reactive protein; GI gastrointestinal; GNB gram-negative bacteria; ISx immunosuppressant; MDR multidrug-resistant; WBC white blood cell.

Discussion
Bacterial Infection represent the most important cause of morbidity after major abdominal surgery [20,21]. Notable, GNB was the predominant pathogens seen in patients who undergoing abdominal surgery.
In the era of increasing prevalence of MDR bacterial infections, MDR GNB is strongly associated with mortality in critically ill surgical patients [22]. Our study investigated the independent risk factors associated with MDR GNB infection after abdominal surgery and highlights MDR GNB infection is associated with a lower 30-day survival compared to non-MDR GNB infection. Our ndings are valuable for clinicians working in highly stressful environment such as SICU and provides useful practical information on prevention and early intervention of MDR GNB infection in SICU patients who received abdominal surgery.
Hasanin et al. [10] analyzed SICU patients who underwent various types of surgery found that 65% of 234 Several studies have been done to determine the risk factors for acquisition of carbapenem-resistant GNB infection [23][24][25]. Longer hospital stay, previous exposure to anti-pseudomonal penicillin, anti-pseudomonal cephalosporins and carbapenems have been reported to be independent risk factors for acquiring carbapenem-resistant GNB (70% cases were glucose non-fermenting GNB) infection [14].  27]. Prior antibiotic use and duration of antibiotic treatment is a concern for those patients before admission to SICU. Likewise, the increase in the pre-ICU length of stay might, however, suggest delay in ICU referral. These scenarios pose the emerging for development of MDR GNB infection.
Previous studies have showed that hyperbilirubinemia is associated with greater severity of illness and poor outcome [28,29]. Fleid et al reported that hyperbilirubinemia in the SICU patients predisposes to infection [30]. Elevation of bilirubin level in ICU patients is usually associated with severe infections such as GNB sepsis or septic shock [31,32]. An observational study conducted in 283 critically ill patients exposed sepsis is one of the most important factors of hyperbilirubinemia [32]. Our study was the rst analysis conducted in critically ill abdominal surgery patients unveiled that hyperbilirubinemia is associated with MDR GNB infection. We believe that hyperbilirubinemia in this series is due to severe MDR GNB sepsis rather than anatomic biliary abnormality. Accordingly, it should judiciously take into account the threat and potential MDR bacterial infection when elevated bilirubin level was found in SICU patients. Further studies to elucidate more information on this respect are warranted.
Several studies have demonstrated that re-exploration of abdomen after the rst operation was associated with prolonged length of ICU stay, higher mortality rate and more frequent wound infections [33][34][35]. The most common complications requiring re-laparotomy were bleeding, sepsis, anastomotic leaks, bowel obstruction and wound dehiscence. We have no information concerning whether the patients received urgent or elective re-operation, but our nding shown that re-exploration is an important risk factor for acquiring MDR bacterial infection. This result is undoubtedly emphasizing the importance of thoroughness at the rst laparotomy.
An important key nding in our study is exposure to anti-pseudo carbapenems and uoroquinolones is associated with the acquisition of MDR bacterial infection. Fluoroquinolones, which are associated with MDR Pseudomonas aeruginosa have also been identi ed as a risk factor for carbapenem-resistant Klebsiella pneumoniae infection [36,37]. Prior carbapenems use also have been identi ed as signi cant independent risk factors for carbapenem-resistant Klebsiella pneumoniae infection [38]. Consistent with our study, antibiotic exposure was associated with an increased risk of developing MDR bacteria but the relationship is a multifaceted issue such as intra-and inter-hospital transmission and antimicrobial resistance in community level. Given the increasingly MDR GNB infections, implementation of antimicrobial stewardship programs in critical care unit is crucial to reduce and optimize antimicrobial therapies.
In our series, the in-hospital mortality rate was as high as 56% in patients with MDR GNB infection. Our ndings exposed the independent risk factors for 30-day mortality after onset of MDR GNB infection were chronic kidney disease, anti-pseudo carbapenems, and inappropriate empirical antimicrobial therapy. Undoubtedly, delay in starting appropriate treatment is associated with increased morbidity and mortality [39,40]. Mosdell et al. shown 480 patients with secondary bacterial peritonitis, patients who received empirical treatment with an appropriate antimicrobial agent had fewer wound infections, abscesses, reoperations, and lower mortality when compared to patients who received inappropriate therapy [39]. While inappropriate therapy certainly has an important impact on clinical outcomes, our series emphasized needs to be considered in the context of other patient risk-factors such as co-morbid conditions. Our study not only highlights the importance of appropriate antimicrobial therapy in improving outcomes, but also underscores that chronic kidney disease is one of key factors that in uence outcomes. Previous studies underscore patients with chronic kidney disease had exceedingly high mortality rates compared with patients without chronic kidney disease in the clinical course of sepsis [41,42]. Our study results emphasize efforts are needed to reduce the negative effects of infections in patients with chronic kidney disease.
In addition, this study illustrating use of anti-pseudo carbapenems is risk factor for subsequent MDR GNB infection and increase 30-day mortality after developed MDR GNB infection. The risk of acquisition of carbapenem-resistant GNB increases with carbapenem exposure [43]. As a result, the emergence of resistant to carbapenems GNB has severely challenged antimicrobial therapy. This indicates the limitation of using anti-pseudo carbapenems as empirical therapeutic options in critically ill surgical patients. The agents of last resort against MDR organisms include the tigecycline, aminoglycosides and polymyxins particularly carbapenem-resistant isolates, but it can be also associated with more signi cant adverse effects (i.e., nephrotoxicity, ototoxicity, and neurotoxicity) [44][45][46][47][48][49][50]. Consideration of therapeutic effectiveness and adverse effects, combination therapy, at least in the empirical phase of treatment, is the higher probability that a MDR isolate will be susceptible to at least one agent in combination regimens and minimize the adverse effects [51,52]. Before the development of novel antimicrobials that could provide clinical e cacy towards MDR organisms, multiple interventions should be employed to control the spreading of MDR organisms including antimicrobial stewardship policies and appropriate infection control measures [13,14]. Implementation of online antimicrobial-stewardship program has been signi cantly reduced antimicrobial consumption and expenditure in the ICU setting [13]. Further study is essential to investigate the effects on reducing the incidence of antimicrobial resistant before and after implementation of antimicrobial-stewardship program.
Several limitations of our study should be noted. First, this study was conducted at a single medical center which might be biased by patient referral patterns. Second, being a retrospective study, power calculation to determine the sample size was not performed, and unavoidable that some data are missing included the interval between time elapsed from diagnosis to surgery and information regarding the treatment of open surgical wound. Further, information about adequate source control was not available. Nevertheless, our study assessed patient who requiring drainage and surgical re-exploration, as source control generally involves drainage of abscesses or infected uid, and debridement of necrotic or infected tissues.

Conclusion
This study has underscored the high prevalence rate of MDR GNB infection and provides important information about the risk factors for the development of MDR GNB infection in patients with abdominal surgery. Risk factors like chronic kidney disease, anti-pseudo carbapenems and inappropriate antibiotics therapy were associated with increased 30-day mortality. Our study highlights the alarming need of multidisciplinary efforts to de ne the optimal strategy for the empirical treatment of patients at the risk for MDR GNB infection.